Developed ML dev software for testing and comparing different sentence embedding models within an Alexa speech recognition model architecture.

Built software to accelerate evaluation process by producing metrics (confusion matrix, accuracy, precision, F1, etc.) in a concise manor, measuring speech recognition architecture performance with different sentence embedders.

Working on project to support customer engagement through AI/ML.

Collected data on transportation options within MA to create heat-maps of MA to visualize price and travel time of possible victims to the closest opioid treatment facility.

Utilized python to scrape opiod and transportation data from internet, and used libraries such as geoplot and GeoPandas to create heat-maps.

Built a service based off a web API integrated with an SDK to serve pre-trained NLP model prediction results to other SurveyMonkey services.

Estimated memory usage reduction of over 70% for NLP models running on server nodes.